CVE-2024-7983 — HIGH (CVSS 7.5) AI Security Vulnerability

Q: Is CVE-2024-7983 actively exploited?

No confirmed active exploitation of CVE-2024-7983 has been reported, but organizations should still patch proactively.

Q: How to fix CVE-2024-7983?

1. PATCH: Upgrade open-webui beyond version 0.3.8 — verify a fixed release is available before upgrading. 2. NETWORK CONTROL (interim): Restrict access to the markdown conversion endpoint (`/api/utils/markdown`) via WAF rule or reverse proxy — block or require authentication on this specific route. 3. RATE LIMITING: Add rate limiting at the reverse proxy level (nginx/Caddy) for all unauthenticated API endpoints. 4. DETECTION: Monitor for requests to the markdown endpoint with unusually large payloads or long processing times; alert on server response time degradation. 5. NETWORK HARDENING: If open-webui is for internal use only, ensure it is not exposed to the public internet — enforce authentication at the network perimeter.

Q: What systems are affected by CVE-2024-7983?

This vulnerability affects the following AI/ML architecture patterns: LLM chat interfaces, self-hosted AI platforms, AI model serving, agent frameworks.

Q: What is the CVSS score for CVE-2024-7983?

CVE-2024-7983 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.41%.

CISO Take

An unauthenticated endpoint in open-webui ≤0.3.8 allows any network-reachable attacker to hang the server by sending a malicious markdown payload, denying LLM access to all users until the request completes. If your team uses open-webui as a self-hosted LLM interface, treat this as a single-request availability kill switch — no credentials required. Patch immediately or firewall the markdown conversion endpoint as an interim control.

Risk Assessment

CVSS 7.5 High, but EPSS of 0.16% indicates limited active exploitation observed. Exploitability is trivially low-barrier: network access + one HTTP request, no authentication or AI/ML knowledge needed. Blast radius is availability-only (C:N/I:N/A:H) — no data exfiltration risk. The primary concern is deployment context: open-webui instances exposed to untrusted networks (internal threat actors, exposed dev environments) face meaningful availability risk. The lack of a patch version listed increases urgency.

Affected Systems

Package	Ecosystem	Vulnerable Range	Patched
open-webui	pip	<= 0.3.8	No patch
135.3K Pushed 8d ago 58% patched ~9d to patch Full package profile →

Do you use open-webui? You're affected.

Severity & Risk

CVSS 3.1

7.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 61% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ CISA SSVC: Public PoC

Composite signal derived from CISA KEV, CISA SSVC, EPSS, trickest/cve, and Nuclei templates.

Attack Surface

AV Network

AC Low

PR None

UI None

S Unchanged

C None

I None

A High

Recommended Action

5 steps

PATCH

Upgrade open-webui beyond version 0.3.8 — verify a fixed release is available before upgrading.
NETWORK CONTROL (interim): Restrict access to the markdown conversion endpoint (/api/utils/markdown) via WAF rule or reverse proxy — block or require authentication on this specific route.
RATE LIMITING

Add rate limiting at the reverse proxy level (nginx/Caddy) for all unauthenticated API endpoints.
DETECTION

Monitor for requests to the markdown endpoint with unusually large payloads or long processing times; alert on server response time degradation.
NETWORK HARDENING

If open-webui is for internal use only, ensure it is not exposed to the public internet — enforce authentication at the network perimeter.

CISA SSVC Assessment

Decision Track*

Exploitation poc

Automatable Yes

Technical Impact partial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

Classification

DoS Auth Bypass API Framework AML.T0029 - Denial of AI Service AML.T0034 - Cost Harvesting AML.T0049 - Exploit Public-Facing Application

Compliance Impact

This CVE is relevant to:

EU AI Act

Art. 9 - Risk Management System

ISO 42001

A.6.2.6 - Security of AI system inputs

NIST AI RMF

MANAGE 2.2 - Mechanisms for sustaining operational resilience

OWASP LLM Top 10

LLM04 - Model Denial of Service

Frequently Asked Questions

What is CVE-2024-7983?

An unauthenticated endpoint in open-webui ≤0.3.8 allows any network-reachable attacker to hang the server by sending a malicious markdown payload, denying LLM access to all users until the request completes. If your team uses open-webui as a self-hosted LLM interface, treat this as a single-request availability kill switch — no credentials required. Patch immediately or firewall the markdown conversion endpoint as an interim control.

Is CVE-2024-7983 actively exploited?

No confirmed active exploitation of CVE-2024-7983 has been reported, but organizations should still patch proactively.

How to fix CVE-2024-7983?

1. PATCH: Upgrade open-webui beyond version 0.3.8 — verify a fixed release is available before upgrading. 2. NETWORK CONTROL (interim): Restrict access to the markdown conversion endpoint (`/api/utils/markdown`) via WAF rule or reverse proxy — block or require authentication on this specific route. 3. RATE LIMITING: Add rate limiting at the reverse proxy level (nginx/Caddy) for all unauthenticated API endpoints. 4. DETECTION: Monitor for requests to the markdown endpoint with unusually large payloads or long processing times; alert on server response time degradation. 5. NETWORK HARDENING: If open-webui is for internal use only, ensure it is not exposed to the public internet — enforce authentication at the network perimeter.

What systems are affected by CVE-2024-7983?

This vulnerability affects the following AI/ML architecture patterns: LLM chat interfaces, self-hosted AI platforms, AI model serving, agent frameworks.

What is the CVSS score for CVE-2024-7983?

CVE-2024-7983 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.41%.

Technical Details

NVD Description

In version 0.3.8 of open-webui, an endpoint for converting markdown to HTML is exposed without authentication. A maliciously crafted markdown payload can cause the server to spend excessive time converting it, leading to a denial of service. The server becomes unresponsive to other requests until the conversion is complete.

Exploitation Scenario

An adversary discovers an open-webui instance via Shodan or passive DNS reconnaissance. Without any credentials, they craft a markdown payload designed to trigger catastrophic backtracking in the regex-based parser (e.g., deeply nested markdown structures, malformed tables, or pathologically complex inline formatting). A single POST to the unauthenticated markdown-to-HTML endpoint causes the server process to peg CPU for an extended period. Legitimate users attempting to interact with the LLM receive timeouts or connection failures for the duration. The attacker can sustain the outage by sending repeated requests, effectively maintaining a denial of service with minimal effort and no authentication. In AI-assisted operations environments (SOC tooling, developer assistants), this disrupts AI-augmented workflows during the outage window.